This paper, available here, provides an overview of the state of economic analysis of unilateral effects in mergers with differentiated products. It discusses both static and dynamic competition.
Section 2 focuses on price competition and discusses the calibration of unilateral effects using diversion-based tools such as upward pricing pressure.
One of the most prominent developments of the past decades was to put closeness of substitution at the heart of unilateral effects analysis. It is well known that market shares can be off the mark in trying to account for consumers’ heterogeneous switching patterns between differentiated products. When robust data is available, it is therefore more sensible to assess competitive overlaps directly via diversion ratios than to rely on market shares as an imperfect proxy. Obtaining an estimate of diversion is feasible in many, though far from all, significant mergers (e.g., through switching data, bidding data, customer surveys, event studies or demand estimation).
While diversion ratios provide a good indication of the relative closeness of substitution between merging companies, they do not capture the absolute intensity of competition faced by firms in the market. A more complete assessment requires going one step further, usually in the form of upward pricing pressure (UPP) assessments. At the heart of the UPP approach is the observation that competing to attract new customers inflicts competitive harm on the other merging firm through the cannibalisation of its business. Post-transaction, merging firms will take this cost of competing into account and thus act less aggressively. This is usually modelled through tools such as the gross upward pricing pressure index (GUPPI), which reflects diverted sales and the financial cost of diversion – i.e. GUPPIs not only capture whom firms compete with but also how much. Anti-competitive effects are more likely if both of these parameters are significant. However, GUPPIs ignore subsequent feedback effects, as the price reactions of firms in the market are iteratively passed through into higher post-merger prices. An ingenious alternative to address this issue is to ask which post-merger subsidy would lead to pre-merger prices. The result is called compensating marginal cost reductions (CMCRs). CMCRs measure the size of marginal cost efficiencies that would be necessary to offset the upward pricing pressure caused by a merger.
Price pressure tools such as GUPPIs and CMCRs have considerable advantages over more traditional forms of analysis. First, they accurately reflect both the intensity of competition in the market and the relative closeness of substitution between the merging parties. Second, they have limited informational requirements that can often be satisfied in significant mergers. Third, they do not require markets to be defined. Fourth, they allow to integrate efficiencies seamlessly into the competitive analysis. Finally, they have simple and intuitive interpretations, representing, respectively, the pre-merger tax that would generate post-merger prices (GUPPIs) and the size of efficiencies needed to offset competitive harm (CMCRs).
How upward pricing pressure impacts prices is determined by the extent to which price pressure is passed through into post-merger prices. This is the realm of merger simulation, which entails the calibration of likely price effects through the use of more specific economic models. There are two broad categories of merger simulation that are used in antitrust practice: calibrated merger simulations and simulations based on demand estimation. Calibrated merger simulations attempt to predict price changes based on simple pre-merger observables such as market shares, diversion ratios and margins. Merger simulation based on demand estimation relies instead on an econometric estimation of the parameters of some model of competition. For this reason, demand estimation tends to have more demanding data requirements.
Practically all of the tools discussed in this section are regularly used by competition agencies. An important question this raises is whether different tools should be viewed as substitutes or as complements. In using these tools, there is generally a trade-off between complexity and precision. More complex methods, such as merger simulation, can give more precise predictions if the underlying model is correctly specified. However, that cannot always be presumed. Simpler methods such as price pressure indices have countervailing advantages in terms of robustness, as they restrict attention to the core of the anticompetitive effect. On balance, this suggests the use of different approaches as complementary evidence to obtain a fuller picture of the potential drivers of competitive effects.
In European case practice, there has arguably been a trend towards linear merger simulation at the expense of other methods described in this article. Although many decisions also reported GUPPIs, CMCRs and IPRs, the Commission’s emphasis in recent years was mostly on calibrated simulation. While the assumption of a particular demand form can be restrictive, the Commission noted that the specific assumption of linear demand is conservative, as other forms of demand, such as log-linear demand, would imply a higher predicted price increase. Recent trends at the CMA appear to have been somewhat different, and to be moving towards GUPPIs. The CMA’s 2017 retail mergers commentary even describes GUPPIs as the most commonly used measure for the quantification of competitive effects.
Section 3 goes on to assess innovation competition.
This section focuses on how agencies evaluate horizontal mergers that may lessen innovation. Earlier innovation literature suggested a complex relation between innovation and competition, and often focused on issues such as innovation and market structure, which are related to but not identical to the question of how mergers affect innovation. In effect, it has become apparent that one cannot generalise the results of those papers easily to a merger context. Recent literature shies away from more general (but also generic) characterisations of innovation, and competition and instead ask the more specific question of what happens in an industry before and after a merger when innovation is an important parameter of competition alongside other choice variables such as prices. Perhaps the most important outcome of this literature has been to re-focus attention on the unilateral effects in innovation rivalry that may result from a merger.
The authors – acknowledging that one of them has taken a strong position in the debate – summarise the current state of play. They consider that two main channels have been identified, absent efficiencies or uninternalised spillovers: the innovation externality channel (which always reduces innovation incentives) and the price coordination channel (which may reinforce or dilute the former). Therefore, to the extent that R&D expenditures spent today are geared at winning sales from rivals tomorrow (or protecting own sales from them), a merger between competitors with significant R&D overlaps can be problematic for competition. While innovation is important, it should not be forgotten that ultimately it is the impact on consumers that needs to be evaluated. Since higher future prices are bad for consumers (all else equal), the price coordination channel will only rarely outweigh the first channel, unless there are strong efficiencies (such as R&D synergies) in addition. However, one should distinguish between situations where one firm’s R&D drives profits away from rivals, and situations where innovation instead increases rivals’ profits. In practice, an important source of such pro-competitive complementarities can be knowledge spillovers. More generally, efficiencies can improve merger effects if they are merger-specific. One such efficiency can be the ability better to organise R&D internally. In short, there is no unique presumption that mergers are bad for innovation and consumers, even if the authors consider that it is reasonable for competition authorities to begin their analysis with the guiding principle that, absent merger-related efficiencies, a horizontal merger is unlikely to have positive effects on innovation incentives.
Section 4 discusses the role of market shares and structural presumptions in differentiated product markets.
Diversion-based methods have considerable advantages over concentration-based measures in differentiated product markets, since they reflect closeness of substitution and market power much more accurately than market shares could ever hope to. Moreover, properly delineating markets requires essentially the same type of information used to analyse competitive effects (e.g., diversion ratios and margins). Yet, market definition produces only a crude in or out decision with these inputs, whereas diversion-based tools reflect the underlying substitution patterns and competitive intensity contained in the data. This is why effects-based methods have considerably advanced the assessment of mergers with differentiated products. Even so, market shares rightfully continue to play an appreciable role in the analysis of differentiated product mergers. This includes the use of safe harbours for low market share mergers and rebuttable presumptions for high market share mergers. Of course, a direct assessment of competitive effects will tend to be preferable to mere concentration analysis when the available data is robust. However, in a world of imperfect information, market data is not always perfect. This militates for considering both types of evidence in an investigation on their respective merits.
Contrary to what is sometimes claimed, structural presumptions for high market share mergers are well grounded in economic theory. In some cases, the link between market shares and market power can be quite direct. A stricter approach towards the acquisitions of firms with high market shares therefore rightly protects the limited level of competition that still exists in highly concentrated markets. Structural presumptions also go some way towards addressing more complex competition concerns, such as impediments to potential competition, innovation competition or coordinated effects. Of course, none of this implies that market shares are an intrinsically more desirable diagnostic tool than effects-based methods. The point here is instead that high (low) market shares establish a plausible prior for (against) competition concerns absent more specific information about the likely effects of a merger. In this way, concentration analysis can play an important role in shaping the standard of proof against which effects-based analysis can be judged.
- Section 5 concludes with an outlook.
We are in the midst of a sharp increase in pricing power across the economy. This secular shift is in large part driven by a reallocation of economic activity towards larger, more profitable oligopolists. In such an environment, vigilant merger control is particularly important to protect the functioning of the competitive process. As the empirical literature has shown, horizontal mergers can cause substantial competitive harm in oligopolistic markets. Thus, it is crucial to determine sensible methods that can detect potentially anticompetitive mergers and distinguish them from benign transactions. This article provided a comparative treatment of different economic tools that assess competitive effects. These tools draw on the key factors that determine competitive outcomes (in particular, who merging firms compete with, and how much) to assess whether or not a transaction is likely to lead to substantial unilateral effects. It is also important to recognise, however, that economic tools are not a panacea. In particular, quantitative predictions should not be interpreted as surgical point-estimates, but as directional measures that reflect the expected strength of underlying incentives. Appropriately applied, quantitative techniques should therefore be embedded in a broader theory of the case that is based on a combination of qualitative and quantitative pieces of evidence to understand the likely competitive implications of a given transaction.
This paper is built around the tenth anniversary of the US Horizontal Merger Guidelines. I stripped most of the references to these Guidelines from this review, since I believe the paper’s most important contribution is to provide an overview of economic thinking on unilateral effects, independently of the content of those Guidelines. This, it does rather well and in a way that even a lawyer can understand.
Overall, the paper strikes me as rather balanced – with the caveats that I am far from an expert on these matters, and that the discussion on innovation competition takes a pro-interventionist slant in the ongoing debate, even if far more measured than earlier papers by these authors, as we shall see in a few weeks.
One thread that runs through the paper, that may not show up in the discussion above but which I must emphasise, is that the tools discussed are designed to illuminate specific aspects of competition, and that other aspects may make adverse effects more or less likely. Quantitative tools should therefore not be used mechanically, but in the context of a coherent theory of harm that combines quantitative with qualitative evidence, including an analysis of potential mitigating factors that may act as constraints on merger effects. I think this is a very important idea to bear in mind when using any tool or piece of evidence in competition cases more generally.